Predictors Of Radiation Pneumonitis Research Articles

Radiomic and Dosiomic Features for the Prediction of Radiation Pneumonitis Across Esophageal Cancer and Lung Cancer.

PurposeThe aim was to investigate the advantages of dosiomic and radiomic features over traditional dose-volume histogram (DVH) features for predicting the development of radiation pneumonitis (RP), to validate the generalizability of dosiomic and radiomic features by using features selected from an esophageal cancer dataset and to use these features with a lung cancer dataset.Materials and MethodsA dataset containing 101 patients with esophageal cancer and 93 patients with lung cancer was included in this study. DVH and dosiomic features were extracted from 3D dose distributions. Radiomic features were extracted from pretreatment CT images. Feature selection was performed using only the esophageal cancer dataset. Four predictive models for RP (DVH, dosiomic, radiomic and dosiomic + radiomic models) were compared on the esophageal cancer dataset. We further used a lung cancer dataset for the external validation of the selected dosiomic and radiomic features from the esophageal cancer dataset. The performance of the predictive models was evaluated by the area under the curve (AUC) of the receiver operating characteristic curve (ROCAUC) and the AUC of the precision recall curve (PRAUC) metrics.ResultThe ROCAUCs and PRAUCs of the DVH, dosiomic, radiomic and dosiomic + radiomic models on esophageal cancer dataset were 0.67 ± 0.11 and 0.75 ± 0.10, 0.71 ± 0.10 and 0.77 ± 0.09, 0.71 ± 0.11 and 0.79 ± 0.09, and 0.75 ± 0.10 and 0.81 ± 0.09, respectively. The predictive performance of the dosiomic- and radiomic-based models was significantly higher than that of the DVH-based model with respect to esophageal cancer. The ROCAUCs and PRAUCs of the DVH, dosiomic, radiomic and dosiomic + radiomic models on the lung cancer dataset were 0.64 ± 0.18 and 0.37 ± 0.20, 0.67 ± 0.17 and 0.37 ± 0.20, 0.67 ± 0.16 and 0.45 ± 0.23, and 0.68 ± 0.16 and 0.44 ± 0.22, respectively. On the lung cancer dataset, the predictive performance of the radiomic and dosiomic + radiomic models was significantly higher than that of the DVH-based model. However, the PRAUC of the dosiomic-based model showed no significant difference relative to the corresponding RP prediction performance on the lung cancer dataset.ConclusionThe results suggested that dosiomic and CT radiomic features could improve RP prediction in thoracic radiotherapy. Dosiomic and radiomic feature knowledge might be transferrable from esophageal cancer to lung cancer.

Dosimetric Risk Factors for Acute Radiation Pneumonitis in Patients With Prior Receipt of Immune Checkpoint Inhibitors.

PurposeDosimetric parameters (e.g., mean lung dose (MLD), V20, and V5) can predict radiation pneumonitis (RP). Constraints thereof were formulated before the era of combined immune checkpoint inhibitors (ICIs) and radiotherapy, which could amplify the RP risk. Dosimetric predictors of acute RP (aRP) in the context of ICIs are urgently needed because no data exist thus far.Methods and MaterialsAll included patients underwent thoracic intensity-modulated radiotherapy, previously received ICIs, and followed-up at least once. Logistic regression models examined predictors of aRP (including a priori evaluation of MLD, V20, and V5), and their discriminative capacity was assessed by receiver operating characteristic analysis.ResultsMedian follow-up of the 40 patients was 5.3 months. Cancers were lung (80%) or esophageal (20%). ICIs were PD-1 (85%) or PD-L1 (15%) inhibitors (median 4 cycles). Patients underwent definitive (n=19), consolidative (n=14), or palliative (n=7) radiotherapy; the median equivalent dose in 2 Gy fractions (EQD2) was 60 Gy (IQR, 51.8-64 Gy). Grades 1-5 aRP occurred in 25%, 17.5%, 15%, 2.5%, and 5%, respectively. The only variables associated with any-grade aRP were V20 (p=0.014) and MLD (p=0.026), and only V20 with grade ≥2 aRP (p=0.035). Neither the number of prior ICI cycles nor the delivery of concurrent systemic therapy significantly associated with aRP risk. Graphs were constructed showing the incrementally increasing risk of aRP based on V20 and MLD (continuous variables).ConclusionsThis is the first study illustrating that V20 and MLD may impact aRP in the setting of prior ICIs. However, these data should not be extrapolated to patients without pre-radiotherapy receipt of prior ICIs, or to evaluate the risk of chronic pulmonary effects. If these results are validated by larger studies with more homogeneous populations, the commonly accepted V20/MLD dose constraints could require revision if utilized in the setting of ICIs.

Radiation pneumonitis in patients with non-small-cell lung cancer receiving chemoradiotherapy and an immune checkpoint inhibitor: a retrospective study

BackgroundImmunotherapy has been administered to many patients with non-small-cell lung cancer (NSCLC). However, only few studies have examined toxicity in patients receiving an immune checkpoint inhibitor (ICI) after concurrent chemoradiotherapy (CCRT). Therefore, we performed a retrospective study to determine factors that predict radiation pneumonitis (RP) in these patients.MethodsWe evaluated the size of the planning target volume, mean lung dose (MLD), and the lung volume receiving more than a threshold radiation dose (VD) in 106 patients. The primary endpoint was RP ≥ grade 2, and toxicity was evaluated.ResultsAfter CCRT, 51/106 patients were treated with ICI. The median follow-up period was 11.5 months (range, 3.0–28.2), and RP ≥ grade 2 occurred in 47 (44.3%) patients: 27 and 20 in the ICI and non-ICI groups, respectively. Among the clinical factors, only the use of ICI was associated with RP (p = 0.043). Four dosimetric variables (MLD, V20, V30, and V40) had prognostic significance in univariate analysis for occurrence of pneumonitis (hazard ratio, p-value; MLD: 2.3, 0.009; V20: 2.9, 0.007; V30: 2.3, 0.004; V40: 2.5, 0.001). Only V20 was a significant risk factor in the non-ICI group, and MLD, V30, and V40 were significant risk factors in the ICI group. The survival and local control rates were superior in the ICI group than in the non-ICI group, but no significance was observed.ConclusionsPatients receiving ICI after definitive CCRT were more likely to develop RP, which may be related to the lung volume receiving high-dose radiation. Therefore, several factors should be carefully considered for patients with NSCLC.

Biological dosiomic features for the prediction of radiation pneumonitis in esophageal cancer patients

ObjectiveThe purpose of this study was to develop a model using dose volume histogram (DVH) and dosiomic features to predict the risk of radiation pneumonitis (RP) in the treatment of esophageal cancer with radiation therapy and to compare the performance of DVH and dosiomic features after adjustment for the effect of fractionation by correcting the dose to the equivalent dose in 2 Gy (EQD2).Materials and methodsDVH features and dosiomic features were extracted from the 3D dose distribution of 101 esophageal cancer patients. The features were extracted with and without correction to EQD2. A predictive model was trained to predict RP grade ≥ 1 by logistic regression with L1 norm regularization. The models were then evaluated by the areas under the receiver operating characteristic curves (AUCs).ResultThe AUCs of both DVH-based models with and without correction of the dose to EQD2 were 0.66 and 0.66, respectively. Both dosiomic-based models with correction of the dose to EQD2 (AUC = 0.70) and without correction of the dose to EQD2 (AUC = 0.71) showed significant improvement in performance when compared to both DVH-based models. There were no significant differences in the performance of the model by correcting the dose to EQD2.ConclusionDosiomic features can improve the performance of the predictive model for RP compared with that obtained with the DVH-based model.

Prediction of radiation pneumonitis with machine learning using 4D-CT based dose-function features.

In this article, we highlight the fundamental importance of the simultaneous use of dose-volume histogram (DVH) and dose-function histogram (DFH) features based on functional images calculated from 4-dimensional computed tomography (4D-CT) and deformable image registration (DIR) in developing a multivariate radiation pneumonitis (RP) prediction model. The patient characteristics, DVH features and DFH features were calculated from functional images by Hounsfield unit (HU) and Jacobian metrics, for an RP grade ≥ 2 multivariate prediction models were computed from 85 non-small cell lung cancer patients. The prediction model is developed using machine learning via a kernel-based support vector machine (SVM) machine. In the patient cohort, 21 of the 85 patients (24.7%) presented with RP grade ≥ 2. The median area under curve (AUC) was 0.58 for the generated 50 prediction models with patient clinical features and DVH features. When HU metric and Jacobian metric DFH features were added, the AUC improved to 0.73 and 0.68, respectively. We conclude that predictive RP models that incorporate DFH features were successfully developed via kernel-based SVM. These results demonstrate that effectiveness of the simultaneous use of DVH features and DFH features calculated from 4D-CT and DIR on functional image-guided radiotherapy.

Machine Learning Classifier Using Pretreatment Factors to Predict Radiation Pneumonitis Following Stereotactic Body Radiation Therapy for Lung Cancer

<h3>Purpose/Objective(s)</h3> Symptomatic radiation pneumonitis (RP) is a common cause of treatment related toxicity in patients receiving radiation therapy for lung cancer. Machine learning techniques have been utilized to determine predictors of RP, but most of these models have used dosimetric features to generate results. The identification of pretreatment factors related to the development of symptomatic RP and implementation of these factors into a machine learning model would be clinically useful to guide treatment decisions, risk estimation, and risk reduction of development of symptomatic RP. <h3>Materials/Methods</h3> We retrospectively reviewed data from 278 lung cancer patients that underwent stereotactic body radiation therapy (SBRT). Pretreatment factors including patient demographics, pulmonary function data, and tumor characteristics were previously analyzed for the development of symptomatic RP, defined as CTCAE v4.0 ≥ Grade 2. A programming environment was used to generate classification machine learning models based on this dataset. Models were tested for accuracy in classification, area-under-the-curve (AUC), and special attention was paid to the false negative rate. The performance of several different algorithms using various combinations of predictors was evaluated including Decision Trees, Discriminant Analysis, Logistic Regression, Naive Bayes Classifiers, Support Vector Machines, Nearest Neighbor Classifiers and Ensemble Classifiers. Each model utilized 5-fold cross-validation to prevent overfitting. <h3>Results</h3> Of the 278 patients receiving SBRT, 42 patients (15.1%) developed symptomatic RP. A RUSBoosted Trees Ensemble Classifier was found to be the most accurate model tested. Features identified as significant risk factors for symptomatic RP in prior univariate and multivariate analyses were found to be the most important predictors in the machine learning models tested. These features included prior radiation treatment to the thorax or lung resection, location of tumor in the right lower lobe, a higher T Stage and age. When this model was used to classify cases of symptomatic RP, it had an AUC of 0.86 and an accuracy of 87%, correctly predicting 36 out of 42 pneumonitis cases. The model had a false negative rate of 14% and false positive rate of 13%. <h3>Conclusion</h3> A machine learning model was developed to identify pretreatment factors predictive of symptomatic RP. The advantage of this model when compared to others with a similar goal is that it depends solely on pretreatment factors, and therefore could be applied in a clinical setting prior to radiation therapy. Next steps include testing with an independent dataset to confirm the model's accuracy and prospective studies to verify the model's clinical utility.

Radiological Prediction Model of Lung Radiation Pneumonitis Based on Dose Line Segmentation

Due to the differences in patients' lung conditions, radiation pneumonia (RP) sometimes occurs even if the lung's dose limitation is met. We are aiming to further predict the occurrence odd of radiation pneumonitis in NSCLC after receiving radiotherapy. Consider based on the dose-volume level to segment the lungs receiving radiation, use radiomic modeling, and evaluate its predictive effectiveness.The study retrospectively enrolled 306 CT scans from 102 NSCLC patients who received IMRT or VMAT from Oct 2015 to Aug 2016. A 0-5 grade RP occurred within 12 months after radiotherapy, considered as the endpoint event. The evaluation of radiation pneumonia in this study case was based on CTCAE 5.0. The whole cohort was divided into two groups, the RP and non-RP. The lung was segmented based on the dose-volume line as V5-10, V10-20, V20-30, V30-40, V40-50, and V50-60 as the region of interest (ROI). Then we register the ROI to the lung of the CT images in the middle and after the radiotherapy. We utilize open source software to extract the image features for each ROI to develop and validate the radiomics model to predict radiation pneumonitis. Two classifier models of random forest (Random forest, RF) and support vector machine (SVM) were used for classifying and learning samples' characteristics to distinguish between the pneumonia-occurring group and the non-occurring group effectively. We randomly divided the sample size into a training and validation group at a 4:1 ratio, and the five-fold cross-validation model (K fold cross-validation) was used for verification. The predictive performance was evaluated using overall accuracy for this triple classification task.The overall incidence of radiation pneumonia was 36%, and the incidence of severe radiation pneumonia (grade 3 and above) was 4%. The support vector machine (SVM) model has the best prediction performance compared to random forest, with an average accuracy rate of 0.72 and an average AUC value of 0.66. Under different dose line volume segmentation, the accuracy of the prediction models in pre-, mid-, and after treatment are 0.724, 0.744, 0.775 (V5-10), 0.736, 0.719, 0.737, (V10-20), 0.728, 0.696, 0.693 (V20-30), 0.671, 0.678, 0.684 (V30-40), 0.747, 0.724, 0.705 (v40-50), 0.799, 0.743, 0.726 (V50-60), respectively. The model constructed in the 50-60Gy area of CT before radiotherapy has the best performance in predicting radiation pneumonia.Radiation pneumonia is a side effect that needs to be paid attention to in thoracic radiotherapy, and its incidence and severity directly affect patients' survival. Traditional radiation pneumonia prediction methods are relatively general. With the development of precise radiotherapy, further accurate radiation pneumonia prediction models are required. Radiomics can become a reliable prediction method for radiation pneumonia.

Prediction of Radiation Pneumonitis Using Genome-Scale Flux Analysis of RNA-Seq Derived From Peripheral Blood

Purpose: Radiation pneumonitis (RP) frequently occurs during a treatment course of chest radiotherapy, which significantly reduces the clinical outcome and efficacy of radiotherapy. The ability to easily predict RP before radiotherapy would allow this disease to be avoided.Methods and Materials: This study recruited 48 lung cancer patients requiring chest radiotherapy. For each participant, RNA sequencing (RNA-Seq) was performed on a peripheral blood sample before radiotherapy. The RNA-Seq data was then integrated into a genome-scale flux analysis to develop an RP scoring system for predicting the probability of occurrence of RP. Meanwhile, the clinical information and radiation dosimetric parameters of this cohort were collected for analysis of any statistical associations between these parameters and RP. A non-parametric rank sum test showed no significant difference between the predicted results from the RP score system and the clinically observed occurrence of RP in this cohort.Results: The results of the univariant analysis suggested that the tumor stage, exposure dose, and bilateral lung dose of V5 and V20 were significantly associated with the occurrence of RP. The results of the multivariant analysis suggested that the exposure doses of V5 and V20 were independent risk factors associated with RP and a level of RP ≥ 2, respectively. Thus, our results indicate that our RP scoring system could be applied to accurately predict the risk of RP before radiotherapy because the scores were highly consistent with the clinically observed occurrence of RP.Conclusion: Compared with the standard statistical methods, this genome-scale flux-based scoring system is more accurate, straightforward, and economical, and could therefore be of great significance when making clinical decisions for chest radiotherapy.

Prediction of radiation pneumonitis after definitive radiotherapy for locally advanced non-small cell lung cancer using multi-region radiomics analysis

To predict grade ≥ 2 radiation pneumonitis (RP) in patients with locally advanced non-small cell lung cancer (NSCLC) using multi-region radiomics analysis. Data from 77 patients with NSCLC who underwent definitive radiotherapy between 2008 and 2018 were analyzed. Radiomic feature extraction from the whole lung (whole-lung radiomics analysis) and imaging- and dosimetric-based segmentation (multi-region radiomics analysis) were performed. Patients with RP grade ≥ 2 or < 2 were classified. Predictors were selected with least absolute shrinkage and selection operator logistic regression and the model was built with neural network classifiers. A total of 49,383 radiomics features per patient image were extracted from the radiotherapy planning computed tomography. We identified 4 features and 13 radiomics features in the whole-lung and multi-region radiomics analysis for classification, respectively. The accuracy and area under the curve (AUC) without the synthetic minority over-sampling technique (SMOTE) were 60.8%, and 0.62 for whole-lung and 80.1%, and 0.84 for multi-region radiomics analysis. These were improved 1.7% for whole-lung and 2.1% for multi-region radiomics analysis with the SMOTE. The developed multi-region radiomics analysis can help predict grade ≥ 2 RP. The radiomics features in the median- and high-dose regions, and the local intensity roughness and variation were important factors in predicting grade ≥ 2 RP.

Abstract 2376: Diagnostic and predictive biomarkers of radiation pneumonitis

Abstract Increased rates of treatment failure in lung cancer patients necessitate treatment with high doses of radiation, which often cause toxicities. However, normal tissue injury in lung may manifest as acute toxicities or as delayed radiation pneumonitis. Identification of early indicators of delayed toxicities will allow timely management and mitigation. In present study, using mouse model, we identified a set of circulating microRNAs as biomarkers that potentially allow early detection of radiation induced lung injury such as inflammation and pneumonitis. Longitudinal analysis of serum samples collected from mice exposed to whole thorax lung irradiation (WTLI) allowed discovery of a panel of evolutionary conserved miRNAs that are potential predictors of radiation pneumonitis. Among these, miR21 and miR29a, miR130a, and miR19a were found peaked between 2-4 weeks after radiation, while miR146a exhibited a peculiar biphasic expression. Some of these changes were found to be associated with concomitant changes in anti- and pro-inflammatory serum cytokines. Several of the biomarkers identified in WTLI model, showed overlapping response in a bleomycin mouse model of acute lung injury. We further tested the translational utility of these biomarkers in single drop blood collected from mice using a validated internal normalizer, miR23a (Yadav et al., Science Translational Medicine, 2020, PMID: 32669422). The response noted for several of the candidate biomarkers identified in mouse model are conserved in specimens collected from human patients who received radiation therapy. In conclusion, our study has identified a panel of microRNA biomarkers with potential to develop as early predictor of pulmonary toxicities in patients receiving therapeutic radiation, and in victims of radiation accidents for timely mitigation. Citation Format: Marshleen Yadav, Joseph Liu, Zahida Qamri, Naduparambil K. Jacob. Diagnostic and predictive biomarkers of radiation pneumonitis [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 2376.

A novel nomogram containing acute radiation esophagitis predicting radiation pneumonitis in thoracic cancer receiving radiotherapy

BackgroundRadiation-induced pneumonitis (RP) is a non-negligible and sometimes life-threatening complication among patients with thoracic radiation. We initially aimed to ascertain the predictive value of acute radiation-induced esophagitis (SARE, grade ≥ 2) to symptomatic RP (SRP, grade ≥ 2) among thoracic cancer patients receiving radiotherapy. Based on that, we established a novel nomogram model to provide individualized risk assessment for SRP.MethodsThoracic cancer patients who were treated with thoracic radiation from Jan 2018 to Jan 2019 in Shandong Cancer Hospital and Institute were enrolled prospectively. All patients were followed up during and after radiotherapy (RT) to observe the development of esophagitis as well as pneumonitis. Variables were analyzed by univariate and multivariate analysis using the logistic regression model, and a nomogram model was established to predict SRP by “R” version 3.6.0.ResultsA total of 123 patients were enrolled (64 esophageal cancer, 57 lung cancer and 2 mediastinal cancer) in this study prospectively. RP grades of 0, 1, 2, 3, 4 and 5 occurred in 29, 57, 31, 0, 3 and 3 patients, respectively. SRP appeared in 37 patients (30.1%). In univariate analysis, SARE was shown to be a significant predictive factor for SRP (P < 0.001), with the sensitivity 91.9% and the negative predictive value 93.5%. The incidence of SRP in different grades of ARE were as follows: Grade 0–1: 6.5%; Grade 2: 36.9%; Grade 3: 80.0%; Grade 4: 100%. Besides that, the dosimetric factors considering total lung mean dose, total lung V5, V20, ipsilateral lung mean dose, ipsilateral lung V5, and mean esophagus dose were correlated with SRP (all P < 0.05) by univariate analysis. The incidence of SRP was significantly higher in patients whose symptoms of RP appeared early. SARE, mean esophagus dose and ipsilateral mean lung dose were still significant in multivariate analysis, and they were included to build a predictive nomogram model for SRP.ConclusionsAs an early index that can reflect the tissue’s radiosensitivity visually, SARE can be used as a predictor for SRP in patients receiving thoracic radiation. And the nomogram containing SARE may be fully applied in future’s clinical work.

The role of peripheral inflammatory cells in predicting radiation pneumonitis

The role of peripheral inflammatory cells in predicting radiation pneumonitis

The role of peripheral inflammatory cells in predicting radiation pneumonitis

The role of peripheral inflammatory cells in predicting radiation pneumonitis

Application of a radiation pneumonitis prediction model in patients with locally advanced lung squamous cell cancer.

The aim of the present study was to establish a new prediction model for radiation pneumonitis (RP) in locally advanced non-small cell lung cancer (LA-NSCLC) patients before and after radiotherapy. The study involved 153 patients. Age, arterial partial oxygen pressure (PO2), forced vital capacity, pulmonary emphysema (PE), subclinical interstitial lung disease (sILD), and dosimetric parameters, such as mean lung dose and percentage of lung volume, and a dose >5/20 Gy (V5/V20), were considered candidate RP predictors. Of the 153 eligible patients, 33 (21.6%) developed RP, 68 had PE (43.8%), and 24 (15.7%) had sILD. Grades 2, 3, and 5 RP were scored in 17 (11.1%), 15 (9.8%), and 1 (0.7%) patient/s, Grade 4 RP was not observed. Grades 1, 2, and 3 PE were scored in 45 (29.4%), 22 (14.4%), and 1 (0.7%) patient/s. Grades 0 and 1 sILD were observed in 129 (84.3%) and 24 (15.7%) patients. Univariate analysis found age, PE, and sILD to be significantly correlated with grade ≥2 RP. Multivariate analysis revealed age >68 years, PE grade >1, and sILD grade ≥1 as independent risk factor for grade ≥2 RP in LA-NSCLC with squamous cell carcinoma (SCC). Finally, a new predictive risk score (PRS) comprised of these factors was developed. The PRS score was 0, 3-5, and 6-11 when the cumulative incidence of grade ≥2 RP was 8.8% (5 patients), 13% (3 patients), and 84.6% (13 patients) (P=<0.001). Age, PE, and sILD could independently and significantly predict RP in LA-NSCLC with SCC.

Multi-institutional dose-segmented dosiomic analysis for predicting radiation pneumonitis after lung stereotactic body radiation therapy.

To predict radiation pneumonitis (RP) grade 2 or worse after lung stereotactic body radiation therapy (SBRT) using dose-based radiomic (dosiomic) features. This multi-institutional study included 247 early-stage nonsmall cell lung cancer patients who underwent SBRT with a prescribed dose of 48-70Gy at an isocenter between June 2009 and March 2016. Ten dose-volume indices (DVIs) were used, including the mean lung dose, internal target volume size, and percentage of entire lung excluding the internal target volume receiving greater than x Gy (x=5, 10, 15, 20, 25, 30, 35, and 40). A total of 6,808 dose-segmented dosiomic features, such as shape, first order, and texture features, were extracted from the dose distribution. Patients were randomly partitioned into two groups: model training (70%) and test datasets (30%) over 100 times. Dosiomic features were converted to z-scores (standardized values) with a mean of zero and a standard deviation (SD) of one to put different variables on the same scale. The feature dimension was reduced using the following methods: interfeature correlation based on Spearman's correlation coefficients and feature importance based on a light gradient boosting machine (LightGBM) feature selection function. Three different models were developed using LightGBM as follows: (a) a model with ten DVIs (DVI model), (b) a model with the selected dosiomic features (dosiomic model), and (c) a model with ten DVIs and selected dosiomic features (hybrid model). Suitable hyperparameters were determined by searching the largest average area under the curve (AUC) value in the receiver operating characteristic curve (ROC-AUC) via stratified fivefold cross-validation. Each of the final three models with the closest the ROC-AUC value to the average ROC-AUC value was applied to the test datasets. The classification performance was evaluated by calculating the ROC-AUC, AUC in the precision-recall curve (PR-AUC), accuracy, precision, recall, and f1-score. The entire process was repeated 100 times with randomization, and 100 individual models were developed for each of the three models. Then the mean value and SD for the 100 random iterations were calculated for each performance metric. Thirty-seven (15.0%) patients developed RP after SBRT. The ROC-AUC and PR-AUC values in the DVI, dosiomic, and hybrid models were 0.660±0.054 and 0.272±0.052, 0.837±0.054 and 0.510±0.115, and 0.846±0.049 and 0.531±0.116, respectively. For each performance metric, the dosiomic and hybrid models outperformed the DVI models (P<0.05). Texture-based dosiomic feature was confirmed as an effective indicator for predicting RP. Our dose-segmented dosiomic approach improved the prediction of the incidence of RP after SBRT.

P04.02 Three Definitions of Normal Lung Volumes for Predicting Radiation Pneumonitis in Advanced NSCLC Treated With Radical Chemoradiation Therapy

Chemoradiotherapy followed by durvalumab is the standard treatment for the patients with unresectable stage III non-small cell lung cancer (NSCLC), but Grade 2 or greater radiation pneumonitis (RP2) is a major obstacle of the use of following consolidative immunotherapy. Although several dosimetric parameters were found to be predictive of RP2, including mean lung dose (MLD) and the lung volume receiving ≥20 Gy (V20), very few researches have noted different definition of normal lung volume may affect the predictive value of the dosimetric parameters, especially in the curative-intent volumetric intensity modulated arc therapy (VMAT) planning, and limited data regarding the onset of RP2 in different time windows. The purpose of this study was to investigate associations between clinical and three definitions of normal lung volumes and RP2. We retrospectively analyzed 117 stage III NSCLC patients treated with chemotherapy concurrent or sequential VMAT from January 2015 to May 2019 in Guangdong Lung Cancer Institute. Patients with palliative dose (<60 Gy), poor performance status (≥2) or <6 months follow-up were excluded. The incidence of radiation pneumonitis in lung cancer was counted every two weeks. Dosimetric parameters including MLD, V20 and the lung volume receiving ≥5 Gy (V5) were calculated based on three definition of normal lung which excluding GTV (Lung-GTV), CTV (Lung-GTV) or PTV (Lung-GTV) from the total bilateral lung volume. The different RP2 prediction ability was assessed by the areas under the curve (AUC) of receiver operating characteristic (ROC) curves. Clinical and dosimetric factors regarding radiation pneumonitis were evaluated using repeated analysis of variance. The median age of patients was 60 years and median follow-up was 1.4 years. 79(67.5%) patients received concurrent cisplatin/etoposide (45,38.5%) or carboplatin/paclitaxel (28,23.9%) and 38(32.5%) patients received sequential carboplatin/paclitaxel (29,24.8%).Grade 2 or more radiation pneumonitis were developed in 31 (26.5%) patients. Of these, 18(58.1%) occurred within 12 weeks, 10(32.3%), 3(9.6)developed between 12 to 24 weeks and after 24 weeks respectively. There is significant heterogeneity in MLD, V20 among the various definition methods (all p<.05), average MLD, V20 were 16.3Gy,28.6%; 15.5Gy,27.4% and 14.3 Gy,25.6% when excluding GTV, CTV and PTV, respectively. The best RP2 prediction performance is found in MLD from Lung-GTV method (AUC = 0.688), which is significantly better (P = 0.002) than Lung-CTV method (AUC = 0.672) and Lung-CTV method (AUC = 0.654). RP2 most commonly developed within 12 weeks after radiotherapy. MLD from the GTV exclusion method may be more accurate in predicting clinically significant radiation pneumonitis while V5, V20 or any of the clinical factors are shows no statistical significance.

The Plasma Levels and Polymorphisms of Vitronectin Predict Radiation Pneumonitis in Patients With Lung Cancer Receiving Thoracic Radiation Therapy

Our previous findings have identified vitronectin (VTN) as a potential biomarker for radiation pneumonitis (RP) through proteomics and molecular mechanism studies. In a recent study, we further explored associations of plasma level and single nucleotide polymorphisms of VTN with the risk of RP in patients with lung cancer receiving radiation therapy. A total of 165 patients with lung cancer were prospectively enrolled with detection of VTN concentration before radiation therapy. VTN reference single nucleotide polymorphisms, rs704 and rs2227721, were genotyped by Taqman probe method. Cox proportional hazard models were performed to identify clinical variables and genotypes associated with the risk of RP on univariate and multivariate analyses, and t tests and analysis of variance were conducted to evaluate the expression level of VTN. The baseline secretion level of VTN in patients with grade ≥3 RP was significantly higher than that in grade <3 RP patients (P < .0001), and elevated levels were observed in patients having the AA genotype compared with GA/GG genotypes of rs704. The VTN rs704 GA/GG and rs2227721 AA/AC genotypes had a significantly lower risk of RP (hazard ratio [HR], 0.448, P = .005; HR, 0.419, P = .008, respectively). In addition, combining cut-off values of mean lung dose (MLD) and VTN plasma level, grade ≥3 RP risk groupings were as follows: high risk: MLD ≥12 Gy and VTN level ≥132 μg/mL (RP rate, 10 of 16 patients, 62.5%); intermediate risk: MLD ≥12 Gy and VTN level <132 μg/mL or MLD <12 Gy and VTN level ≥132 μg/mL (8 of 70 patients, 11.4%); and low risk: MLD <12 Gy and VTN level <132 μg/mL (1 of 79 patients, 1.3%) (P < .0001). Among patients receiving radiation therapy, relatively high plasma levels of VTN before radiation therapy were associated with the higher incidence of RP, and VTN rs704 and rs2227721 each had a significant effect on predicting RP risk. Combining VTN concentration with MLD appeared to facilitate stratification of patients with lung cancer who received radiation therapy into low-, intermediate-, and high-risk RP groups. This study indicated that VTN may serve as a blood biomarker for susceptibility to RP in patients with lung cancer.

Dosimetric Factors and Radiomics Features Within Different Regions of Interest in Planning CT Images for Improving the Prediction of Radiation Pneumonitis.

This study aimed to establish machine learning models using dosimetric factors and radiomics features within 5 regions of interest (ROIs) in treatment planning computed tomography images to improve the prediction of symptomatic radiation pneumonitis (RP) (grade ≥2). This study retrospectively collected data on 79 patients with lung cancer (25 RP ≥2) who underwent chemoradiotherapy between 2015 and 2018. We defined 5 ROIs in planning computed tomography images: gross tumor volume (GTV), planning tumor volume (PTV), PTV-GTV, total lung (TL)-GTV, and TL-PTV. We calculated the mean dose, V5, V10, V20, and V30 within TL-GTV and TL-PTV and the mean dose within the other ROIs. A total of 1924 radiomics features were extracted from all 5 ROIs. We selected the best predictors for classifying 2 groups of patients using a sequential backward elimination support vector machine model. A permutation test was used to assess its statistical significance (P < .05). The best predictors for symptomatic RP were the combination of 11 radiomics features, 5 dosimetric factors, age, and T stage, achieving an area under the curve (AUC) of 0.94 (95% confidence interval [CI], 0.85-1) (accuracy, 90%; sensitivity, 80% [95% CI, 44%-96%]; specificity, 95% [95% CI, 73%-100%]; P = 8 × 10-4). The clinical characteristics, dosimetric factors, and their combination showed limited predictive power (accuracy, 63.3%, 70%, and 70%; AUC [95% CI]: 0.73 [0.54-0.92], 0.53 [0.31-0.75], and 0.72 [0.51-0.92], respectively). The radiomics features of PTV-GTV and TL-PTV outperformed those of the other ROIs (accuracy, 76.7% and 76.7%; AUC [95% CI]: 0.82 [0.65-0.99] and 0.80 [0.59-1], respectively). Combining dosimetric factors and radiomics features within different ROIs can improve the prediction of symptomatic RP. Our results can help physicians adjust the radiation dose distribution of the dose-sensitive lungs and target volumes based on personalized RP estimates.

Prediction of Radiation Pneumonitis With Machine Learning in Stage III Lung Cancer: A Pilot Study.

Background:Radiation pneumonitis (RP) is a dose-limiting toxicity in lung cancer radiotherapy (RT). As risk factors in the development of RP, patient and tumor characteristics, dosimetric parameters, and treatment features are intertwined, and it is not always possible to associate RP with a single parameter. This study aimed to determine the algorithm that most accurately predicted RP development with machine learning.Methods:Of the 197 cases diagnosed with stage III lung cancer and underwent RT and chemotherapy between 2014 and 2020, 193 were evaluated. The CTCAE 5.0 grading system was used for the RP evaluation. Synthetic minority oversampling technique was used to create a balanced data set. Logistic regression, artificial neural networks, eXtreme Gradient Boosting (XGB), Support Vector Machines, Random Forest, Gaussian Naive Bayes and Light Gradient Boosting Machine algorithms were used. After the correlation analysis, a permutation-based method was utilized for as a variable selection.Results:RP was seen in 51 of the 193 cases. Parameters affecting RP were determined as, total(t)V5, ipsilateral lung Dmax, contralateral lung Dmax, total lung Dmax, gross tumor volume, number of chemotherapy cycles before RT, tumor size, lymph node localization and asbestos exposure. LGBM was found to be the algorithm that best predicted RP at 85% accuracy (confidence interval: 0.73-0.96), 97% sensitivity, and 50% specificity.Conclusion:When the clinical and dosimetric parameters were evaluated together, the LGBM algorithm had the highest accuracy in predicting RP. However, in order to use this algorithm in clinical practice, it is necessary to increase data diversity and the number of patients by sharing data between centers.

Radiomics-Based Radiation Pneumonitis Prediction on Pretreatment Planning Computed Tomography Images Prior to Lung Cancer Stereotactic Body Radiation Therapy

Radiomics-Based Radiation Pneumonitis Prediction on Pretreatment Planning Computed Tomography Images Prior to Lung Cancer Stereotactic Body Radiation Therapy